1. EVOLUTIONARY
COMPUTING
M.K.H.Gunasekara
AS2010377
Department of Computer Science
University of Sri Jayewardenepura

2. OVERVIEW
 Introduction
 Problem
 Assumptions
 Player’s Performance
 Solution
2

3. INTRODUCTION
 Genetic algorithms (GAs) are numerical
optimization algorithms inspired by both natural
selection and natural genetics
 Introduced by John Holland
 Genetic Algorithm promises results in NP-Complete
problems
3

4. PROBLEM
 A cricket team for Sri Lanka has to be selected via
genetic algorithms without the human intervention.
The team must comprise a captain, a wicket-
keeper, one all-rounder, four batsmen, two spinners
and two fast bowlers. Assume that the Sri Lankan
cricket board has a pool of players of captains,
wicket-keepers, all-Rounders, batsmen, spinners
and fast bowlers. The GA's task is to pick the best
cricket team (11 players) out of the pool.
4

5. ASSUMPTIONS
 Pools of Players are mutually exclusive
 Size of the Pool is not static
 There is record of each player’s performance
 There is no correlation between players
5

6. PLAYER’S PERFORMANCES
 Like ICC Ratings
6

7. CHROMOSOME REPRESENTATION
Captain Wicket
Keeper
All
Rounder
Batsman 1 Batsman 2 Batsman 3 Batsman 4 Fast
Bowler 1
Fast
Bowler 2
Spin
Bowler 1
Spin
Bowler 2
7
 Chromosome should be fixed size
 Size - 11
 11-genes

8. BIT REPRESENTATION
 In java we can represent each bit by using object
references
 We can use Player’s pool index
8

9. FITNESS FUNCTION
 Summation of the player’s performance
 Fitness= Captain * 3 +
WicketKeeper +
AllRounder +
Batsman(1) * 2 +
Batsman(2) * 2 +
Batsman(3) * 2 +
Batsman(4) * 2 +
FBowler(1) +
FBowler(2) +
SBowler(1) +
SBowler(2) +
 Add Penalty for invalid chromosomes 9

10. SELECTION METHOD
 Tournament Selection method
 It always neglect invalid chromosomes
 This method is more efficient and leads to an
optimal solution.
 Winner from a larger tournament will on average
have a higher fitness than the winner of a smaller
tournament
10

11. CROSSOVER OPERATOR
 Single point Crossover
11
Crossover Position

12. HOW TO TACKLE INVALID OFF SPRINGS
 Calculate the fitness of two off springs
 Then insert fittest off spring in to the new population
 Eg:
12
1 23 46 55 65 75 90 2 5 8 9
1 23 46 55 80 83 65 8 7 9 10
1 23 46 55 65 75 65 8 7 9 10
1 23 46 55 80 83 90 2 5 8 9

13. MUTATION OPERATOR
 Flipping Method
 Flipping of a bit involves changing value to
available different value in the pool on a mutation
chromosome generated
13
1 23 46 55 65 75 90 2 5 8 9
Mutation point
All Rounder Pool
46

14. REPLACEMENT METHOD
 Used Steady State Update method with replacing
worst member in the current generation
 Increase the Genetic drift
14

15. ELITE PRESERVING OPERATOR
 Reserve current fittest team into next generation
 In this way, not only best solutions of the current
population pass into the next generation, but they
also participate with other members of the
population in creating offspring
15

16. TUNING GENETIC ALGORITHM PARAMETERS
 Change Tournament size
 Change Crossover probability
 Change Mutation probability
16

17. IMPLEMENTATION
17

18. THANK YOU
18